This program has concentrated increasingly on the structural and genomic basis of neoplastic progression in the human GI mucosa. It is hypothesized that better understanding of the biology of carcinogenesis will facilitate early diagnosis by providing clues that point to those patients most susceptible to cancer even before they have clear histologic evidence of neoplastic disease. The genomic, proliferative, clonal, and morphologic abnormalities will bc studied prospectively in three gastrointestinal diseases with a potentially malignant phase: Barrett's esophagus, the precursors of sporadic colon cancer, and ulcerative colitis. Adenocarcinoma of the esophagus is almost always a complication of Barrett's esophagus rarely discovered in time for cure. Carcinoma of the colon is the second most frequent cause of cancer death in the USA, curable if diagnosed early, but fatal if diagnosed late. Carcinoma in long-standing ulcerative colitis is rarely curable unless diagnosed well before symptoms appear. Vigorous long-term surveillance can diagnose these cancers early, when surgically curable, but this is costly and time-consuming. Because many patients with these three premalignant conditions will never develop cancer, the problem is in seeking those individuals at highest risk in whom close surveillance is rewarding. The specific aims in all three projects are similar: to identify early steps in the progression to cancer by the study of genomic, proliferative, clonal, and morphological abnormalities retrospectively and prospectively using flow cytometry, cytogenetics, analysis of clonality, molecular genetics, and histopathology. Tissue will come from well-characterize cohorts of patients who have had repeated and extensive endoscopic biopsy and whose operative specimens will be also studied meticulously.